local_dev.md

Local Development

The suite has a python script in the root of the respository, named vision-zero, which can be used to replicate and run the Vision Zero database and graphQL API locally.

See the repo README for quick start instructions for using this tool.

The vision-zero program is a light wrapper around the functionality provided by docker compose. By inspecting the docker-compose.yml file, you can find the definitions of the services in the stack, and you can use the docker compose command to turn up, stop, and attach terminals to the running containers and execute on-off commands. This can provide you access to containers to install nodejs libraries, use postgres' supporting programs (psql, pg_dump) and other lower level utilities.

vision-zero command auto-completion

The vision-zero application is able to generate auto-completion scripts via the shtab python library. For example, zsh users may use the following to enable this feature. bash and csh users will have similar steps to follow particular to their shell of choice.

mkdir ~/.zsh_completion_functions;
chmod g-w,o-w ~/.zsh_completion_functions;
cd $WHEREVER_YOU_HAVE_VZ_CHECKED_OUT;
source ./venv/bin/active;
./vision-zero -s zsh | tee ~/.zsh_completion_functions/_vision-zero

Examples of vision-zero commands

Note: There is a flag which ends up being observed for any of the following commands which start the postgres database:

-r / --ram-disk will cause the database to back its "storage" on a RAM disk instead of non-volatile storage. This has the upside of being much faster as there is essentially no limit to the IOPS available to the database, but the data won't be able to survive a restart and will require being replicate-db'd back into place.

The default is to use the disk in the host to back the database, which is the operation our team is most familiar with, so if you don't need or want the RAM disk configuration, you can ignore this option.

vision-zero build

Rebuild the stack's images based on the Dockerfiles found in the repository. They are built with the --no-cache flag which will make the build process slower, but avoid any stale image layers that have inadvertently cached out-of-date apt resource lists.

vision-zero db-up & vision-zero db-down

Start and stop the postgres database

vision-zero graphql-engine-up & vision-zero graphql-engine-down

Start and stop the Hasura graphql-engine software as well as the database

vision-zero api-up & vision-zero api-down

Start and stop the Vision Zero CR3 / User API

vision-zero local-stack-up & vision-zero local-stack-down

Start and stop the DB, graphql-engine, and the API at once

vision-zero psql

Start a psql postgreSQL client connected to your local database

vision-zero tools-shell

Start a bash shell on a machine with supporting tooling

vision-zero stop

Stop the stack

vision-zero replicate-db

• Download a snapshot of the production database
• Store the file in ./database/snapshots/visionzero-{date}-{with|without}-change-log.sql
• Drop local vision_zero database
• Create and repopulate the database from the snapshot

Note: the -c / --include-change-log-data flag can be used to opt to include the data of past change log events. The schema is created either way. Note: the -f / --filename flag can be optionally used to point to a specific data dump .sql file to use to restore.

The way the snapshots are dated means that one will only end up downloading one copy of the data per-day, both in the with and without change log data.

vision-zero dump-local-db

• pg_dump the current local database
• Stores the file in `./database/dumps/visionzero-{date}-{time}.sql

vision-zero remove-snapshots

Remove snapshot files. This can be done to save space and clean up old snapshots, but it's also useful to cause a new copy of the day's data to be downloaded if an upstream change is made.